So, had a column up this morning about a hot topic in election forecasting: How to think about Democrats' massive fundraising advantage in races for Congress. 53eig.ht/2NS0LYW
One point that I think people miss: the Dems' fundraising numbers are pretty good but not **that** good. It's that Dems' fundraising numbers are pretty good *and* the GOP ones are bad. Moreover, GOP fundraising is shallow. Far fewer of their candidates will hit the $1m mark. 
This leads to an unprecedented 2:1 D fundraising edge in swing districts. Usually even when a party had an overall fundraising edge, it doesn't translate so lopsidedly in swing districts. This time it does, even though a lot of the swing districts are red and have GOP incumbents.
I empathize with the case for sort of ignoring the fundraising numbers—the case basically being, they're so good for Ds/bad for Rs that they seem outlier-ish and we don't know what to do with them. But I don't think that case is prudent or wise. You can't just ignore outliers.
Overall, the fundraising numbers create a lot of upside *and* downside risk for both parties, relative to what models that use them say.
The risk may not be symmetric either. Personally, I think the D seat distribution may be more right-skewed than what our model shows. e.g., maybe the model is right that Ds win 35-40 on average, but the median should be lower (30-35?), and the tail where they win 55+ fatter.
